Amps
[Top] [All Lists]

[Amps] parasitic oscillation techniques

To: amps@contesting.com
Subject: [Amps] parasitic oscillation techniques
From: John Lyles <jtml@losalamos.com>
Reply-to: jtml@vla.com
Date: Sun, 23 Aug 2009 10:15:42 -0600
List-post: <amps@contesting.com">mailto:amps@contesting.com>
The discussion of they what, why and how of parasitic suppression has 
always been controversial on this forum, especially when AG6K was here 
with his own retrofit kits and QST articles. I think there has been a 
huge volume already said for a dozen years on this, and I will add my 
own two cents worth, based on experience, not what I read or heard. I'll 
first add to Lon's explanation of the dominant parasitic mode in many 
3-500Z designs (and other medium power amateur tubes). The circuit 
inductance from the anode cap through the first shunt tuning capacitor 
tends to have enough L by accident (not by design) that it easily 
resonates with the stray C from the tube and the tuning capacitor C. 
This was discussed at length in the out-of-print book RF HEATING TUBES, 
from Mr. Dietrich of Philips Electronic in Eindoven, along with models 
showing how it happens. This one is easy to find with even a grid dip 
meter, but a network analyzer demonstrates its presence too. When we 
design for HF, we are looking at component values with uH and many pF. 
At VHF, it only may take nH and pF to allow sufficient impedance rise. 
This is why layout, lead length, lead diameter, are all important to 
good amplifier stability. Also, these hard-to-quantify strays are what 
makes it seem like hit-or-miss design to dampen elusive parasitic 
resonances. Once this happens, and the tube still has gain with enough 
feedback inside, you have a candidate for power oscillation or other odd 
behavior in the 100's of MHz. This seems to be the reason that certain 
tubes are more prone to the problem when circuited at HF.

As a power amplifier designer of commercial/industrial circuits with 
large tubes, for nearly 30 years, I have had my own personal battles 
with elusive demons in amplifiers and power oscillators. In some 
circuits, the tendency for acting up has been minimal. But precautions 
have always been taken to watch where stray inductances are, and to 
measure, sweep the circuits BEFORE applying HV. I have made use of 
nichrome as a lower Q wire, for various 'suppressors' before I heard of 
their application in SB220s. When I shared this info with Rich Measures 
15 years ago. But I am not an advocate of application of this technique 
in every tube amplifier, as it all depends on the individual circuit 
values and the requirements for power, frequency.

>         The self resonance that causes most of the problems in the SB-220 is 
> around 90 to 105 MHz  It is easily measured, both in frequency and "Q".  In 
> the TL-922A there is also a high UHF self resonance that will vaporize 1/2 
> (only one side) of the double-sided twenty meter fixed contact on the band 
> switch.  Relate the size of one side of the contact to frequency and you 
> will get the frequency of the self resonance.  It is interesting to note 
> that the original TL-922 (not the 922A) dose not have this problem.  This is 
> because of a very slight, physical, re-design of the final compartment of 
> the later TL-922A that was sold on the American market.   Good parasitic 
> suppressors will de-tune/attenuate these self resonances, hopefully to a 
> point where they will be harmless.
.....
>  73 de Lon, K5JV

As Carl mentioned here, there are other factors in amplifiers that can 
also cause RF instabilities, including the B+ choke, and VHF/UHF 
cavity-type resonances. These latter forms can be quite destructive and 
troublesome, especially when one is trying to design a VHF amplifier 
using coaxial or radial transmission line geometry. In my early career, 
I was able to stabilize a commercial amplifier that used a 4CX3500A 
tetrode in a coaxial output circuit by using a combination of nichrome 
wire L and stray C alone. This little LC tab was applied at the right 
place, to effectively quiet down a nervous amplifier that would have 
otherwise been a commercial failure. In another application years later, 
I did a somewhat thorough study of the application of nichrome, steel 
and copper wire, to large B+ chokes in 27 MHz power oscillators (30 kW 
CW). Not only was the  conductor compared, but also the geometry of the 
chokes, length and width, pitch, using a Hp Q meter followed up with 
power tests. The biggest problem there was that the resistive alloy was 
so lossy that it was glowing dull orange due to the plate current alone 
(5 amps DC). Then there are designs that never seemed to need parasitic 
suppressors, remembering one 8877 amplifier at 5 MHz that I built.

Presently am finishing a large amplifier circuit design that has 
occupied me for a number of years as a part-time endeavor, now a full 
press project that has a commercial deadline. Due to the large geometry 
of the power tube and the need for it to develop 3 MW of VHF power, 
there is a significant money and time being allocated for suppression of 
parasitic cavity modes, even before the thing has been turned on. One 
missed high Q mode may destroy the $200K device. Big tubes (that require 
cranes and hoists for lifting) are prone to supporting UHF and L-band 
oscillations due to circular path geometry inside the radial beam tube 
itself. This one will not be easily stopped using lossy wire, instead 
using lossy bulk material (Eccosorb) in waveguide-style mode dampeners.

So I guess my point here is that these things - suppressors - are not 
artforms, but are practical engineered devices, that have their place in 
power tube circuitry. It is unfortunate that their discussion among hams 
(and among some non-ham high power circuit designers - I have been there 
too!) tends to degenerate to the level of unexplained phenomena like 
UFOs. We have powerful tools that we can not use to demonstrate these 
problems in circuits. What we don't have is good knowledge and models of 
the tubes themselves, as to how they are easily tickled into laughter 
(unwanted oscillation).

73
John
K5PRO



_______________________________________________
Amps mailing list
Amps@contesting.com
http://lists.contesting.com/mailman/listinfo/amps

<Prev in Thread] Current Thread [Next in Thread>